breath noise

[HopHead]

Well, I got another tricky one for you guys. I have an LP I am restoring that contains an awful lot of high frequency information. An annoying side effect of this is the breath intake of the vocalist which comes across as pretty overwhelming at certain points.

I have tried like, everything with limited success. The 'best' results to date have been with the CNF, attenuating just one blue point around the 12-13 mhz mark. This does a perfect job of cutting his breath but I guess because of the type of music [country with yodleing] and the fact that I cannot [or do not know how]to 'isolate' this noise on the waveform, there are artifacts created here and there which makes some of the music sound unreal.

I would sure appreciate any help.

Tony

BTW - Happy New Year to everyone!

[greg_m]

Re: therat

I think maybe you're using the wrong filter. The CNF is for *continuous* noise, like tape hiss, etc.

You are trying to filter out *occasional* noise (breath intake) so I think something else might work better. Maybe just using the paragraphic EQ, and dipping out the frequencies you find, would work better.

Of course if you dip out a certain HF band on the entire file, it's going to remove other things at other times ... not just the breathing. You could also highlight just the brief part of the file where the breath noise occurs, and filter just that part.

You might also look at some threads on "de-essing" to remove sibilants; it sounds as if what you want to do might be similar to that. In addition to this forum, look in the newer Live/Five forum... some of the topics there might be applicable for you.

Personally, I would just quit listening to yodelling... [img]images/icons/wink.gif[/img]

Good luck, and Happy New Year!

[HopHead]

Re: therat

Hi Greg

hahaha well - there's no accounting for taste huh?

Anyway thanks for your help but I tried, like every other filter and combination thereof but the CNF was the only one that did any sort of decent job.

I guess the real prob is - isolating the 'noise' which also overlays the music. But don't get me wrong, the results were quite acceptable with just a very small hint of wierdness here and there and not in every song either.
Its just that I felt there should be a better [read more professional]way.

[Craig Maier]

Re: therat

Have you tried to take a fingerprint of the "breath intake" and then used the CNF in spectral subtraction mode? This could work. Find a place where just the "breath intake" exists and sample it. Then use the attenuation control to tune it for the best results.

[Craig Maier]

Re: therat

Sorry - I forgot that Millennium does not have spectral subtraction mode.

[HopHead]

Re: therat

Hi again Craig,

Is the inclusion of this mode planned for Millenium's CNF or should I upgrade?

Sounds like something I should have!

Cheers to you & family.

[Craig Maier]

Re: therat

The spectral subtraction mode in the CNF is standard in the DC 5 and the Live/Forensics versions. We have no plans to retrofit the Millennium product with the new CNF features.